Broken rpm setup, wrong GCC?

[ggamba]

I have both stable and experimental repositories enabled here. This morning I was testing Yumie and, seeing that a newer version of "yum" package was available, I installed it
The program reported that the update was ok. Then I selected "rpm" package and chose "Update". Unfortunately I got an "unrecoverable error" message and now nor yum or rpm work from command line. In popuplog.os2 I get lot of these errors:

02-06-2015 16:11:49 SYS2070 PID 0077 TID 0001 Slot 0094
C:\USR\BIN\RPM.EXE
RPMIO->GCC1._udivdi3
127

---

02-06-2015 16:11:49 SYS2070 PID 0077 TID 0001 Slot 0094
C:\USR\BIN\RPM.EXE
RPM->GCC1._register_frame_table
127

---

02-06-2015 16:11:49 SYS2070 PID 0077 TID 0001 Slot 0094
C:\USR\BIN\RPM.EXE
RPMBUILD->GCC1._register_frame_table
127

I understand that probably Yumie did something wrong, and I'll report that in the appropriate place, but now I need some help to fix the above error with GCC :-)

Thanks
Gabriele

[Silvan Scherrer]

@Gabriele: it's most probably not yumie's fault. it looks like this rpm is broken or something. we are investigating it atm.
Do you eventually remember which packages got installed and which updaed?

I sent you some dll's to test. With these your setup should be ok again.

[ggamba]

Hi Silvan,

I just remember that I selected "rpm-python" and Yumie proposed also "rpm" and "rpm-libs". Then update started but exited due to the unrecoverable error.

Your dll's fixed the problem with yum not running, but now it is firefox/thunderbird that don't run :-)

02-06-2015 17:12:48 SYS2070 PID 0054 TID 0001 Slot 006f
C:\PROGRAMS\FIREFOX_BETA2\FIREFOX.EXE
STDCPP6->GCC1.98
182

I tried to update through yum (an updated GCC was available on stable), but "yum update" ended with:

Killed by SIGSEGV
pid=0x0055 ppid=0x0054 tid=0x0001 slot=0x0076 pri=0x0200 mc=0x0001
C:\USR\BIN\PYTHON2.7.EXE
GCC1 0:00007a51
cs:eip=005b:1f157a51 ss:esp=0053:0011e890 ebp=0011e8b8

ds=0053 es=0053 fs=150b gs=0000 efl=00210206

eax=00000000 ebx=00130501 ecx=fffffff3 edx=00000000 edi=00000003 esi=1d7540dd
Process dumping was disabled, use DUMPPROC / PROCDUMP to enable it.

[Yuri Dario]

Did you previously install gcc-4.9.2.1-1.x (or -2) packages? they were built using ordinals and probably you only updated runtime...
get the above dll here

​http://rpm.netlabs.org/release/00/zip/gcc-4_9_2_1-3_oc00.zip

[ggamba]

Thanks Yuri, with the above DLLs now Firefox and Thunderbird work fine! Almost everything is ok again, just "yum update" is still broken. When I run it, I get:

Upgrade      10 Packages

Dimensione totale del download: 30 M
Procedere [s/N]: s
Download dei pacchetti:
(1/10): ash-0.0.0-11.oc00 100% |=========================|  81 kB    00:00
(2/10): file-libs-5.04-7. 100% |=========================| 344 kB    00:01
(3/10): gcc-4.9.2.1-3.oc0  99% |======================== |  29 MB    00:00 ETA
Killed by SIGSEGV
pid=0x006f ppid=0x006e tid=0x0001 slot=0x0092 pri=0x0200 mc=0x0001
C:\USR\BIN\PYTHON2.7.EXE
GCC1 0:00007a51
cs:eip=005b:1f107a51      ss:esp=0053:0011d690      ebp=0011d6b8
 ds=0053      es=0053      fs=150b      gs=0000     efl=00210206
eax=00000000 ebx=00130501 ecx=fffffff3 edx=00000000 edi=00000003 esi=0fad40dd
Process dumping was disabled, use DUMPPROC / PROCDUMP to enable it.

In Popuplog.os2 I see that there is still a problem with GCC1:

02-06-2015  22:10:56  SYS3175  PID 006f  TID 0001  Slot 0092
C:\USR\BIN\PYTHON2.7.EXE
c0000005
1f107a51
P1=00000002  P2=0fad40ed  P3=XXXXXXXX  P4=XXXXXXXX  
EAX=0fae027c  EBX=0fae029c  ECX=00000000  EDX=00000000
ESI=0fae027c  EDI=00000000  
DS=0053  DSACC=d0f3  DSLIM=5fffffff  
ES=0053  ESACC=d0f3  ESLIM=5fffffff  
FS=150b  FSACC=00f3  FSLIM=00000030
GS=0000  GSACC=****  GSLIM=********
CS:EIP=005b:1f1015d7  CSACC=d0df  CSLIM=5fffffff
SS:ESP=0053:0012fd10  SSACC=d0f3  SSLIM=5fffffff
EBP=0012fd28  FLG=00210202

GCC1.DLL 0001:00007a51

[Yuri Dario]

try also this one

​http://rpm.netlabs.org/release/00/zip/gettext-0_18_1_1-7_oc00.zip

[ggamba]

Do I need everything inside gettext-0_18_1_1-7_oc00.zip?

[Yuri Dario]

just dlls until you fix the system.

[Dave Yeo]

I've had similar errors which often were fixed by a reboot. Seems that some DLLs are unlocked during the update process and left in memory and when an updated program (ash, yum etc) attempt to use gcc1 sys2070 errors are generated.
Both Yum and Yummie need to suggest a reboot on certain updates.

[ggamba]

Hi Yuri,

I had to copy all the DLL, then the python error went away. New problem with "yum update":

Upgrade      11 Packages

Dimensione totale del download: 60 M
Procedere [s/N]: s
Download dei pacchetti:
(1/11): file-libs-5.04-7. 100% |=========================| 344 kB    00:02
(2/11): gcc-4.9.2.1-3.oc0  99% |======================== |  29 MB    00:00 ETA e
rror: rpmts_HdrFromFdno: headerRead failed: hdr blob(109460): BAD, read returned
 109457
(2/11): gcc-4.9.2.1-3.oc0 100% |=========================|  29 MB    00:41


Could not open local rpm file: /@unixroot/var/cache/yum/netlabs-rel/packages/gcc
-4.9.2.1-3.oc00.pentium4.rpm: RPM Error opening Package

I already tried deleting all cache (from Yumie) and rebooting, but the error with the above package persists.

[Yuri Dario]

Please check also RPM files

​http://rpm.netlabs.org/release/00/zip/rpm-4_8_1-22_oc00.zip

[Yuri Dario]

I suggest to check all recent files into ​http://rpm.netlabs.org/release/00/zip and compare DLLS with your existing files.

[ggamba]

​http://rpm.netlabs.org/release/00/zip/rpm-4_8_1-22_oc00.zip did not help. If I try to run rpm from commandline (4OS2), I get the following message:

SYS0123: A file name or volume label contains an incorrect character. "MAGIC"

Any hint?

[Silvan Scherrer]

do a yum reinstall file-libs

[ggamba]

Hi Silvan, "yum reinstall file-libs" did not help. In the end, I got \etc, \usr and \var from a backup, run a "yum update" and now everything seems ok again.

I tried to compare the "broken" directories with the "right" ones, but there are too many differences.

[Silvan Scherrer]

most probably you updated from a wrong or not synced repo. This can happen when files are moved from exp to release and the scheduler did not pick/rebuild the db.

As all seems to work ok now, all rpm's seem good. So I close this ticket.